LBT Innovations brings AI to pathology labs – a healthcare investment worth a (pee)k

  • LBT Innovations has two products MicroStrek and APAS that use AI to analyse cultures
  • APAS is FDA approved and 3 times faster than a manual scientist
  • New product WoundVue will use the same algorithm to assess wounds

Adelaide-based business LBT Innovations is bringing artificial intelligence to the analysis of culture plate processing with two products, MicroStreak and APAS (automated plate assessment system), stepping in for microbiologists.

Brent Barnes, Chief Executive Officer & Managing Director of LBT, says that the APAS technology uses intelligent imaging and machine learning as part of its automation process. The machine will be able to process 200 plates an hour and will be at least three times faster than that of a scientist or a microbiologist.

In 2013 LBT signed a joint venture with Zurich-based Hettich AG Switzerland to commercialise the technology and production of the product will move from Adelaide to Germany after the rollout of product.

The FDA has approved APAS and a trial of the product will begin in St Vincent’s hospital in Melbourne.

Brent Barnes tells Alan Kohler the stock price has settled back down after reacting to the FDA approval and St Vincent’s trial and the company will start making money in 2018.


You’ve got two products, MicroStreak and APAS. They seem to be pretty similar. Is it really a suite of products that automates the back office of pathologies?

They are similar in some respects in that they do deliver automation in the culture plate processing, but they’re quite different with respect to the actual application. MicroStreak is a technology that automates the inoculation and streaking of a specimen. That’s the first stage of the culture plate workflow. It then goes to an incubator and it comes out of an incubator, and then a scientist or microbiologist reads the plate and determines if there’s any significant colony growth.

The second instrument that we’re bringing to market, our automated plate assessment system, or APAS as you explained, essentially does what a scientist or a microbiologist does, reads, interprets the plate to determine if there’s any significant colony growth that occurs or not. It delivers the same outcome with relation to providing automation, but from an application and technology perspective, they’re actually quite different. The APAS technology uses intelligent imaging and machine learning as part of its automation process to essentially replace that of a scientist.

Which of them gets commercialised first?

MicroStreak actually had been commercialised many years ago. We licenced that technology to a large diagnostic company, bioMérieux, in 2009, and that product was brought to market – PREVI Isola was the commercial name. There are some 450 to 500 instruments that had been sold globally under that licencing distribution arrangement, but in 2015, that arrangement was renegotiated and LBT retained the rights back to MicroStreak and our core technology and received about seven and a half million dollars as part of that renegotiation.

That allowed us to reinvest into APAS and really develop the intelligent imaging and machine learning algorithms, which is the second product that we’re bringing to market. We’re going to have that instrument in St. Vincent’s Hospital in Melbourne in September, and that’s something I announced only a couple of weeks ago. That’ll be the first placement as part of a trial site, and then at the end of this year, we’ll plan on starting to sell the product.

In fact, you announced in October that APAS had achieved FDA approval in the United States, and the share price spiked from 19¢ to 87¢. Bang! Then it’s come all the way back down again. Is that just the market doing its thing or have you seized the opportunity to raise some cash and issue some shares?

Alan, it’s probably a mixture of the above. I think the FDA clearance was one that was a great milestone. We actually submitted a 510(k) de novo submission. A de novo means that there’s no predicate device that exists in the US, so there’s no other competing technology that does the automation and the interpretation that APAS does. We got the clearance on the 10th of October of last year, which was great, and the market responded in the subsequent weeks with a huge amount of volume and interest.

That’s all positive but I think some of the challenges that we had at the time as an organisation was knowing that we’re actually still at least 12 months away from having an instrument available for sale. Our regulatory approach was one where we’ve developed a standalone, manual loading, plate reading instrument, which uses our Core Technology. We wanted to de-risk it from a regulatory pathway perspective because there is no instrument which does what we’re doing. We didn’t want to spend a lot of money developing an instrument that the FDA wouldn’t clear. We essentially went through a de-risking process, developed a manual loading instrument, and we did a 10,000 patient clinical trial, which was really on the basis of what the FDA reviewed very critically and it ended up getting cleared, that was a global first.

From a market perspective and the way that we announced it and the share price and the like was all very positive. But perhaps people were buying into the stock not really recognising that we still had a bit of a journey to go as it related to the engineering development of bringing to market an automated system which moved plates around and delivers the level of automation that a pathology lab gets the benefit from. That’s what the APAS Independence instrument actually will do. 200 plates an hour or it will be at least three times faster than that of a scientist or a microbiologist doing the same thing. That’s the exciting part of where we’re at now. We de-risked it from a US perspective, from a Europe perspective, and from a TJ perspective here in Australia. We’re in the latter stages of finishing off the engineering work, going to have an instrument at St. Vincent’s, like I mentioned, in a couple of months with first sales starting in 2018, but planning to ship for our first order at the very end of this year.

In fact, the FDA clearance was just the start in a way. I suppose the market is now looking to see what the results out of St. Vincent’s are. Is that right?

Yes, absolutely, Alan. You’re spot on. It was a big milestone to get the clarity to say,  this is when we’re going to go into St. Vincent and it’s a global first. The market’s going to be keen on two key milestones in the next few months. The first is what the outcomes of those trials are at St. Vincent’s and the second one will be who’s going to distribute our product in larger markets such as North America and Europe. I expect us to be able to make some announcements around that around the September timeframe.

Haven’t you already joint-ventured APAS with a German business?

Yes. We’re bringing APAS Independence to market through a joint venture, and the JV company, Clever Culture Systems, is owned by LBT and also by Hettich, the German company. CCS is essentially the vehicle by which LBT are bringing the technology to market. And LBT owns half of that joint venture company. That means all of the costs with relation to the development are shared, and also, the revenues that come from the sales are also shared throughout our JV agreement, so they’re not a distributor. They’re essentially just the commercial arm or vehicle by which we’ll be bringing APAS Independence to market.

Separately to that, you’re negotiating with distributors, are you?

That’s right. We’ve just come back from a couple of key conferences. The two largest microbiology conferences of the year, one being in Vienna in April, the European Microbiology Conference. There was about 12,000 people who attend this conference, and then just recently in June, the American Society of Microbiology meeting. That was in New Orleans. We took our instrument, the APAS Independence, and had a booth. This was a fully working instrument, so participants were able to actually see plates going through the instrument, photos being captured, seeing the automation occur in terms of the reading and the interpretation of the plates.

It was a fantastic opportunity to get both the market ready, so the end users, the lab directors and microbiologists in the pathology labs, but also an opportunity to talk to potential distributors who’d be looking to add the APAS Independence as part of their portfolio of products that they’re currently selling. To that extent, we are in a due diligence set of processes with a number of potential distributors. Like I said, I hope to finalise some of those things and announce that to the market in around the September timeframe.

Do you think you’ll start making money next year?

Absolutely. 2018 is when we start generating revenue. Essentially, we are moving into a reoccurring revenue stage now. The short answer is definitely yes. The longer answer is this is a journey and we’ll be very targeted in the breadth of countries that we’ll go after first. Australia is our launch market, and that’s really important. LBT are going direct as the distributor here, and that’s because all of the engineering and both of the software and the mechanical parts of the instrument are all here in Australia. Plus, it’s a generation one instrument.

We do expect there will be improvements, and so being able to respond to the customer feedback, understanding what that is very quickly, and being able to bake that into sustaining engineering and improvements to the instrument are really important. We can support it in a very close and reliable fashion. The other thing that we’re doing is we’re building up expertise here to support our distributors in the bigger market, we need to ensure that we have service teams, engineering teams to be able to train our distributors in Europe and in the US to ensure that we’re arming them with all of the information that they need to be able to be successful in those larger markets.

What are you going to charge for it?

The end user price will be around US$300,000 and that will be a once-off fee for the instrument. There’ll be another fee for the software. Really, the software is what drives this instrument. That’s where the intelligence is.

Is the software on a subscription basis?

Correct, that’s a recurring piece of revenue. That will be an annual licence fee. You’ve got your once-off fee for the instrument, US$300,000. You then go and have an annual fee for the software licence. I haven’t disclosed how much that is other than saying it’s a material software licence. The reason I’m not disclosing it is because there’s no other competing product, whereby the software is so meaningful in terms of delivery and value. As we get closer to selling the product at the end of this year, I’ll be able to disclose that, but it’s a material piece of recurring revenue that will be part of the selling process.

What are you telling the pathology labs their return on investment is going to be?

We’ve done some cost utility studies and we’ll plan to publish those over the coming months, but we’re targeting one and a half to two years. The market feedback we’ve had is that typically, they’re looking at a two and a half to three-year payback, so we’re looking to try to bring that forward into one and a half to two years. Of course, that depends on the size of the lab and how many plates that they process. The reimbursement model here in Australia and in larger markets such as the US and the UK, Germany, they get reimbursed per plate. So really, it’s a factor of the number of plates that they process allows them to generate more revenue.

Are you going to be making the machines in Adelaide where you’re based or are you bringing them in from somewhere else?

We’ll be manufacturing them initially with our engineering partner, a company called Planet Innovation, and they’re based in Melbourne. They’re doing all of our engineering work and they’ve built the first set of instruments that we’ve got available. We think for the first maybe 12 months that they’ll manufacture. Our long-term manufacture will be with our joint venture partner Hettich who are based in Germany. They manufacture centrifuge. It was always part of our deal to bring that in house I guess within the JV, but that transfer to manufacturer is one that you don’t want to rush into, especially because we do assume that there’s going to be some changes. So we want to make sure we stabilise our manufacturing processes, ensure that we have the ability to ramp up manufacturing if the demand is much higher than we think, and then we’ll transfer across to our JV partner Hettich who are based in Germany.

What’s your margin on the units?

I can’t disclose how much the margins are, but essentially from how the flow works, we’ve got a distributor, they’re going to have a margin selling the instrument. I can’t explain exactly what that is, but typically, distributors are used to getting a margin of somewhere between 20% to 40%, and then the profit of that comes through the JV and then to each of the JV partners on essentially a shared basis.

I presume the ongoing software fee is all margin, or mostly margin.

That is the nice thing with software, is that once you’ve developed it and you’ve got it out there, there’s really no further development work that’s needed. We will be enhancing the number of plates that we support. One of the things with training the algorithm through machine learning is that we need to train it using different media types. The FDA clearance was for urine, which was great, but there’s a whole bunch of different media that people use and there’s variations in that media that we need to make sure that we train into the algorithm. The core software platform will be developed and will exist, but there will be some minor improvements that we’ll need to continually offer as more media is introduced to the market.

The FDA clearance was just for urine. Do you have to have extra clearances for other media?

Good question. Yes. The short answer is yes. The FDA clearance was for urine, for urinary tract infection as an example. As we go into looking at other specimen types for infection control, for example, and looking at blood samples and other things, we’ll need to submit what’s called a 510(k) supplement, and that’s a much shorter process. The tough thing to get through, which was what we got through in October, was the de novo submission, and that, we’ve now set a predicate. What happens when you’ve got a predicate within FDA is that you go through this 510(k) supplement process, which is really a much more straightforward process than something in for the first time.

I know that you’ve just appointed an artificial intelligence expert to your board, Matthew MichalewiczExplain to us why that is. I hadn’t realised when I started researching your business that actually, what you’re doing is artificial intelligence.

That’s right. The appointment of Matt is really fantastic for this organisation. Really, what this company is all about is our machine learning algorithm using artificial intelligence and our imaging platform. We saw an opportunity to bring someone of the board and that’s got a huge amount of experience around AI, starting out companies and selling companies, or in this space for many, many years. Having someone with that technical computer systems artificial intelligence understanding is a fantastic addition at the board composition level. That was a really helpful appointment.

This company, when you think about LBT, what exists as a platform technology is around our imaging and around these smart algorithms that we’re able to train. To that extent, we have got another product called WoundVue. WoundVue is one that we’re using the same core foundations of this algorithm that we applied for APAS that we’ve trained for looking at chronic wounds. We’ve actually developed a prototype device and we’ve taken photos of a whole set of different wounds. The instrument is able to look at things like your surface area, your volume, your depth, as well as automating the tissue type, so it’s able to identify if there’s slough or granulation that occurs.

That’s really helpful because the current interventions looking at chronic wounds is extremely manual and subjective. You have different nurses and caregivers going out to homes or even within the clinic looking at wounds and making their own assessment, and they’re using very manual processes to do that. By having a device which automates that, you take a photo of it, it’s able to tell you how big the wound is through the surface area, tell you how deep the wound is, and give you that volume, as well as then automating the types of tissues or the tissue types of the wound and able to make an assessment whether the wound is getting better or worse, take that subjectivity out and makes it very objective.

It’s a really good example of how we’re utilising our core platform technology around the imaging and the algorithms in a completely different clinical context. We’re moving from agar plates and looking at specimens and specimen growth on an agar plate which is very defined and symmetrical to a chronic wound which can be very different in terms of how it looks and its depth and its size. The first stage is essentially getting this prototype device out and again utilising that to take more and more images.

Was this technology and the algorithms all developed within LBT in Adelaide or did you buy the technology?

We’ve been working on this for a number of years. Probably around five years. It started through a collaboration with the Australian Centre for Visual Technologies, ACVT, and they’re part of the University of Adelaide. The guy who invented the algorithms of APAS, he’s come on staff to LBT. He was part of the university and he’s moved across, so the inventor is now part of our staff.

What’s his name?

Rhys Hill is his name. He’s our research director here at LBT, but he had long tenure at the University of Adelaide, and so it was great to be able to bring him across and obviously, great for the university. It allowed us to work in that way as well, but we have a longstanding partnership with ACVT, and really, that’s where the foundations of this core platform technology came from. The person who developed it is now on staff with LBT, but we’re still working collaboratively with WoundVue as an example with ACVT. That collaboration, and I’ve got a team of people around this kind of machine learning AI type of methodologies that they have available, but we’ve also brought people on staff, which is helpful.

Was there any one person who dominated the funding of it in the early stages and now dominates the share register or not?

Our share registry is really retail investors and directors in the top 20. We have bioMérieux, which is the big French diagnostic company, as our number one shareholder. They have that 7%. Really, they’re on there through the history of this MicroStreak deal that we had, but really, there’s no strategic conversations happening in any way. We don’t have any arrangement or agreement with bioMérieux anymore, and so that’s just sitting there as our number one shareholder. Other than that, it’s sophisticated investors that sit on our top 20 and we don’t have any significant institutional investors at this time.

Well, it’s been really interesting talking to you, Brent. Congratulations on everything that you’ve done. It’s been fantastic.

Thank you very much, Alan. I appreciate it.